As technology for detecting arcs on an image, a technique for extracting edges and performing Hough transform has been known. The Hough transform is a feature extraction technique that is applicable to detection of graphics in various shapes. With the Hough transform, however, the processing time and the memory consumption increase according to the number of parameters involved in the transform. For detecting an ellipse, since at least four parameters, i.e., coordinates (X, Y) and radii (lateral radius, longitudinal radius) are used, the amount of processing time for the ellipse detection is large and the amount of memory consumption is also large.
Accordingly, technologies for detecting arcs of an ellipse without use of the Hough transform are available. For example, as technology for detecting circular arcs, a technology for detecting circular arcs on the basis of graphic thinning an broken-line approximation is available (e.g., Japanese Unexamined Patent Application Publication No. 07-104954, 63-213087, and 01-180678). Also, as technology for extracting a closed area including an ellipse or the like, a technology for extracting a closed area surrounded by line segments or curves on a map is available (e.g., Japanese Unexamined Patent Application Publication No. 2006-293809).